refactor: rename primitive → numeric in extraction layer

Extraction-layer spec describes numeric types exclusively (bounds,
float_bits). Frees 'primitive' for the broader system-level taxonomy.

Renames:
- PrimitiveSpec → NumericSpec
- primitive_extraction.py → numeric_extraction.py
- extract_primitives → extract_numerics
- partition_primitive_and_geometry_names → partition_numeric_and_geometry_types

partition_numeric_and_geometry_types moved from numeric_extraction to
pipeline — it discovers both numeric and geometry types, so it didn't
belong in a module scoped to numeric extraction.

Renderer function names and output constants unchanged — those describe
rendered output, not the extraction concept.

Signed-off-by: Seth Fitzsimmons <sethfitz@amazon.com>

Author: sethfitz

packages/overture-schema-codegen/README.md

@@ -55,7 +55,7 @@ the same pipeline.
 function -- a single iterative loop that peels NewType, Annotated, Union, and container
 wrappers, accumulating constraints tagged with the NewType that contributed them.
 Domain-specific extractors (`model_extraction`, `union_extraction`, `enum_extraction`,
-`newtype_extraction`, `primitive_extraction`) call `analyze_type()` for field types and
+`newtype_extraction`, `numeric_extraction`) call `analyze_type()` for field types and
 produce spec dataclasses.
 
 **Output Layout** determines what artifacts to generate and where they go. Supplementary

packages/overture-schema-codegen/docs/design.md

@@ -144,7 +144,7 @@ Extraction is split by entity kind:
 - `extraction/enum_extraction.py`: Enum class -> `EnumSpec`
 - `extraction/newtype_extraction.py`: NewType -> `NewTypeSpec`
 - `extraction/union_extraction.py`: Discriminated union alias -> `UnionSpec`
-- `extraction/primitive_extraction.py`: Numeric primitives -> `PrimitiveSpec`
+- `extraction/numeric_extraction.py`: Numeric types -> `NumericSpec`
 
 Each calls `analyze_type()` for field types. Tree expansion (`expand_model_tree()`)
 walks MODEL-kind fields to populate nested model references, with a shared cache and

packages/overture-schema-codegen/docs/walkthrough.md

@@ -230,11 +230,11 @@ dispatch, and example loading all operate on `FeatureSpec` without knowing which
 concrete type they hold.
 
 **EnumSpec** and **EnumMemberSpec** serve enums. **NewTypeSpec** serves NewTypes.
-**PrimitiveSpec** serves numeric primitives with an `Interval` for bounds and optional
+**NumericSpec** serves numeric primitives with an `Interval` for bounds and optional
 `float_bits`.
 
 **SupplementarySpec** is the union type alias `EnumSpec | NewTypeSpec | ModelSpec` --
-the set of non-feature types that need their own output pages. `PrimitiveSpec` and
+the set of non-feature types that need their own output pages. `NumericSpec` and
 geometry types are excluded because they render on aggregate pages rather than
 individual ones.
 
@@ -363,10 +363,10 @@ member for single-value `Literal` fields on the discriminator.
 
 ### Primitive extraction
 
-`partition_primitive_and_geometry_names` reads a module's `__all__` exports. NewType
+`partition_numeric_and_geometry_types` reads a module's `__all__` exports. NewType
 exports are numeric primitives; non-constraint class exports are geometry types.
 
-`extract_primitives` builds `PrimitiveSpec` objects. For each primitive name it resolves
+`extract_numerics` builds `NumericSpec` objects. For each primitive name it resolves
 the object from the module, calls `extract_newtype` for the type analysis, then extracts
 numeric bounds from constraints. `extract_numeric_bounds` scans constraint objects for
 `ge`/`gt`/`le`/`lt` attributes and packs them into an `Interval`.
@@ -713,7 +713,7 @@ two NewType layers and an `Annotated` layer, producing a `TypeInfo` with
 `FieldSpec.model` references. The shared cache ensures sub-models referenced by multiple
 features (like `Sources`) extract once. Union-kind fields skip inline expansion.
 
-**Layout.** `partition_primitive_and_geometry_names` reads the system module's exports.
+**Layout.** `partition_numeric_and_geometry_types` reads the system module's exports.
 `collect_all_supplementary_types` walks Segment's expanded fields and discovers
 referenced enums (like `Subtype`), semantic NewTypes (like `Id`, `Sources`), and
 sub-models. The walk follows `FieldSpec.model` references down the tree, and for

packages/overture-schema-codegen/src/overture/schema/codegen/extraction/field_constraints.py

@@ -28,7 +28,7 @@
 # annotated_types constraint attribute (Ge, Gt, Le, Lt, Interval) to its
 # mathematical symbol for prose rendering.
 #
-# primitive_extraction.py has its own _BOUND_ATTRS for numeric extraction. The
+# numeric_extraction.py has its own _BOUND_ATTRS for numeric extraction. The
 # duplication is deliberate: these modules use the same attribute names for
 # unrelated purposes (display formatting vs. numeric bound extraction), and
 # coupling them for four string literals adds a dependency without value.

…degen/extraction/primitive_extraction.py …codegen/extraction/numeric_extraction.pypackages/overture-schema-codegen/src/overture/schema/codegen/extraction/primitive_extraction.py renamed to packages/overture-schema-codegen/src/overture/schema/codegen/extraction/numeric_extraction.py packages/overture-schema-codegen/src/overture/schema/codegen/extraction/primitive_extraction.py renamed to packages/overture-schema-codegen/src/overture/schema/codegen/extraction/numeric_extraction.py

@@ -1,23 +1,22 @@
-"""Primitive extraction and partitioning."""
+"""Numeric type extraction."""
 
 from annotated_types import Interval
 
 from .docstring import first_docstring_line
 from .newtype_extraction import extract_newtype
-from .specs import PrimitiveSpec, TypeIdentity
-from .type_analyzer import TypeInfo, is_newtype
+from .specs import NumericSpec, TypeIdentity
+from .type_analyzer import TypeInfo
 
 __all__ = [
     "extract_numeric_bounds",
-    "extract_primitives",
-    "partition_primitive_and_geometry_names",
+    "extract_numerics",
 ]
 
 
 # Bound attribute names on annotated_types constraint objects (Ge, Gt, Le,
 # Lt, Interval) used for numeric bound extraction.
 #
-# field_constraint_description.py has its own _BOUND_OPS for display formatting.
+# field_constraints.py has its own _BOUND_OPS for display formatting.
 # The duplication is deliberate: these modules use the same attribute names
 # for unrelated purposes (numeric extraction vs. prose rendering), and
 # coupling them for four string literals adds a dependency without value.
@@ -41,18 +40,18 @@ def extract_numeric_bounds(type_info: TypeInfo) -> Interval:
     return Interval(**found)
 
 
-def extract_primitives(
-    primitive_ids: list[TypeIdentity],
-) -> list[PrimitiveSpec]:
-    """Extract specifications for numeric primitive types."""
-    specs: list[PrimitiveSpec] = []
-    for tid in primitive_ids:
+def extract_numerics(
+    numeric_ids: list[TypeIdentity],
+) -> list[NumericSpec]:
+    """Extract specifications for numeric types."""
+    specs: list[NumericSpec] = []
+    for tid in numeric_ids:
         newtype_spec = extract_newtype(tid.obj)
         bounds = extract_numeric_bounds(newtype_spec.type_info)
         description = first_docstring_line(getattr(tid.obj, "__doc__", None))
         float_bits = _extract_float_bits(tid.name)
         specs.append(
-            PrimitiveSpec(
+            NumericSpec(
                 name=tid.name,
                 description=description,
                 bounds=bounds,
@@ -71,25 +70,3 @@ def extract_primitives(
 def _extract_float_bits(name: str) -> int | None:
     """Extract bit width from a float type name like 'float32'."""
     return _FLOAT_BITS.get(name)
-
-
-def partition_primitive_and_geometry_names(
-    primitive_module: object,
-) -> tuple[list[TypeIdentity], list[TypeIdentity]]:
-    """Discover primitive and geometry types from a module's exports.
-
-    NewType exports are numeric primitives.
-    Non-constraint class/enum exports are geometry types.
-    """
-    module_all: list[str] = getattr(primitive_module, "__all__", [])
-    primitives: list[TypeIdentity] = []
-    geometries: list[TypeIdentity] = []
-
-    for name in module_all:
-        obj = getattr(primitive_module, name)
-        if is_newtype(obj):
-            primitives.append(TypeIdentity(obj, name))
-        elif isinstance(obj, type) and not name.endswith("Constraint"):
-            geometries.append(TypeIdentity(obj, name))
-
-    return primitives, geometries

packages/overture-schema-codegen/src/overture/schema/codegen/extraction/specs.py

@@ -21,7 +21,7 @@
     "FieldSpec",
     "ModelSpec",
     "NewTypeSpec",
-    "PrimitiveSpec",
+    "NumericSpec",
     "PydanticTypeSpec",
     "SupplementarySpec",
     "TypeIdentity",
@@ -189,8 +189,8 @@ class NewTypeSpec(_SourceTypeIdentityMixin):
 
 
 @dataclass
-class PrimitiveSpec:
-    """Extracted specification for a numeric primitive type."""
+class NumericSpec:
+    """Extracted specification for a numeric type."""
 
     name: str
     description: str | None
@@ -219,7 +219,7 @@ def docs_url(self) -> str:
 SupplementarySpec = EnumSpec | NewTypeSpec | ModelSpec | PydanticTypeSpec
 """Non-feature types referenced by feature models.
 
-Excludes PrimitiveSpec and geometry types, which are extracted
+Excludes NumericSpec and geometry types, which are extracted
 separately via dedicated functions.
 """
 

packages/overture-schema-codegen/src/overture/schema/codegen/markdown/path_assignment.py

@@ -31,7 +31,7 @@
 def build_placement_registry(
     feature_specs: Sequence[FeatureSpec],
     all_specs: dict[TypeIdentity, SupplementarySpec],
-    primitive_names: list[TypeIdentity],
+    numeric_names: list[TypeIdentity],
     geometry_names: list[TypeIdentity],
     schema_root: str,
 ) -> dict[TypeIdentity, PurePosixPath]:
@@ -41,7 +41,7 @@ def build_placement_registry(
     the source Python module path relative to schema_root.
     """
     registry: dict[TypeIdentity, PurePosixPath] = _aggregate_page_entries(
-        primitive_names, geometry_names
+        numeric_names, geometry_names
     )
 
     feature_dirs: set[PurePosixPath] = set()
@@ -81,12 +81,12 @@ def resolve_output_path(
 
 
 def _aggregate_page_entries(
-    primitive_names: list[TypeIdentity],
+    numeric_names: list[TypeIdentity],
     geometry_names: list[TypeIdentity],
 ) -> dict[TypeIdentity, PurePosixPath]:
     """Pre-populate registry entries for types documented on aggregate pages."""
     entries: dict[TypeIdentity, PurePosixPath] = dict.fromkeys(
-        primitive_names, PRIMITIVES_PAGE
+        numeric_names, PRIMITIVES_PAGE
     )
     entries.update(dict.fromkeys(geometry_names, GEOMETRY_PAGE))
     return entries

packages/overture-schema-codegen/src/overture/schema/codegen/markdown/pipeline.py

@@ -15,10 +15,7 @@
 
 from ..extraction.examples import ExampleRecord, load_examples
 from ..extraction.model_extraction import expand_model_tree
-from ..extraction.primitive_extraction import (
-    extract_primitives,
-    partition_primitive_and_geometry_names,
-)
+from ..extraction.numeric_extraction import extract_numerics
 from ..extraction.specs import (
     EnumSpec,
     FeatureSpec,
@@ -29,6 +26,7 @@
     TypeIdentity,
     UnionSpec,
 )
+from ..extraction.type_analyzer import is_newtype
 from ..layout.type_collection import collect_all_supplementary_types
 from .link_computation import LinkContext
 from .path_assignment import (
@@ -47,7 +45,11 @@
 )
 from .reverse_references import UsedByEntry, compute_reverse_references
 
-__all__ = ["RenderedPage", "generate_markdown_pages"]
+__all__ = [
+    "RenderedPage",
+    "generate_markdown_pages",
+    "partition_numeric_and_geometry_types",
+]
 
 
 @dataclass(frozen=True, slots=True)
@@ -109,6 +111,28 @@ def _render_supplement(
     return RenderedPage(content=content, path=output_path)
 
 
+def partition_numeric_and_geometry_types(
+    types_module: object,
+) -> tuple[list[TypeIdentity], list[TypeIdentity]]:
+    """Discover numeric and geometry types from a module's exports.
+
+    NewType exports are numeric types.
+    Non-constraint class/enum exports are geometry types.
+    """
+    module_all: list[str] = getattr(types_module, "__all__", [])
+    numerics: list[TypeIdentity] = []
+    geometries: list[TypeIdentity] = []
+
+    for name in module_all:
+        obj = getattr(types_module, name)
+        if is_newtype(obj):
+            numerics.append(TypeIdentity(obj, name))
+        elif isinstance(obj, type) and not name.endswith("Constraint"):
+            geometries.append(TypeIdentity(obj, name))
+
+    return numerics, geometries
+
+
 def generate_markdown_pages(
     feature_specs: Sequence[FeatureSpec],
     schema_root: str,
@@ -123,12 +147,12 @@ def generate_markdown_pages(
     for spec in feature_specs:
         expand_model_tree(spec, cache)
 
-    primitive_names, geometry_names = partition_primitive_and_geometry_names(
+    numeric_names, geometry_names = partition_numeric_and_geometry_types(
         _system_primitive
     )
     all_specs = collect_all_supplementary_types(feature_specs)
     registry = build_placement_registry(
-        feature_specs, all_specs, primitive_names, geometry_names, schema_root
+        feature_specs, all_specs, numeric_names, geometry_names, schema_root
     )
 
     reverse_refs = compute_reverse_references(feature_specs, all_specs)
@@ -148,7 +172,7 @@ def generate_markdown_pages(
 
     pages.append(
         RenderedPage(
-            content=render_primitives_from_specs(extract_primitives(primitive_names)),
+            content=render_primitives_from_specs(extract_numerics(numeric_names)),
             path=PRIMITIVES_PAGE,
         )
     )

packages/overture-schema-codegen/src/overture/schema/codegen/markdown/renderer.py

@@ -23,7 +23,7 @@
     FieldSpec,
     ModelSpec,
     NewTypeSpec,
-    PrimitiveSpec,
+    NumericSpec,
     PydanticTypeSpec,
     TypeIdentity,
     UnionSpec,
@@ -587,8 +587,8 @@ def _bit_width_key(name: str) -> tuple[str, int]:
     return (prefix, int(digits) if digits else 0)
 
 
-def render_primitives_from_specs(specs: list[PrimitiveSpec]) -> str:
-    """Render the primitives.md page from pre-extracted PrimitiveSpecs."""
+def render_primitives_from_specs(specs: list[NumericSpec]) -> str:
+    """Render the primitives.md page from pre-extracted NumericSpecs."""
     template = _get_jinja_env().get_template("primitives.md.jinja2")
 
     signed_ints: list[dict[str, str | None]] = []

packages/overture-schema-codegen/tests/conftest.py

@@ -5,11 +5,11 @@
 from click.testing import CliRunner
 from codegen_test_support import find_model_class
 from overture.schema.codegen.extraction.model_extraction import extract_model
-from overture.schema.codegen.extraction.primitive_extraction import (
-    extract_primitives,
-    partition_primitive_and_geometry_names,
-)
+from overture.schema.codegen.extraction.numeric_extraction import extract_numerics
 from overture.schema.codegen.extraction.specs import ModelSpec
+from overture.schema.codegen.markdown.pipeline import (
+    partition_numeric_and_geometry_types,
+)
 from overture.schema.codegen.markdown.renderer import (
     render_geometry_from_values,
     render_primitives_from_specs,
@@ -72,8 +72,8 @@ def division_class(all_discovered_models: dict) -> type[BaseModel]:
 @pytest.fixture(scope="module")
 def primitives_markdown() -> str:
     """Render the primitives.md page from the system primitive module."""
-    primitive_names, _ = partition_primitive_and_geometry_names(_system_primitive)
-    return render_primitives_from_specs(extract_primitives(primitive_names))
+    numeric_names, _ = partition_numeric_and_geometry_types(_system_primitive)
+    return render_primitives_from_specs(extract_numerics(numeric_names))
 
 
 @pytest.fixture(scope="module")